Mentioned

How to check if ChatGPT recommends your brand (free, 10 minutes)

The short answer: open a fresh ChatGPT chat with web search on and ask the five questions a buyer would ask before choosing a product like yours. Don't use your brand name. Then ask the same five questions in Perplexity. For each answer, write down whether you were named, whether you were the pick, who was picked instead, and which pages were cited. That's 10 answers, and it takes about 10 minutes.

The rest of this guide covers how to make those 10 minutes count, and what to do with what you find.

Before you start: mentioned is not the same as recommended

An AI answer can name your brand without recommending it. "X is another option" is a mention. "I'd tell them to use X" is a recommendation. In our own runs, a brand was often named as an alternative while a competitor got the pick. If you only search the answer for your name, you'll count both the same way and miss the real result.

So you're tracking two things: were we named? and were we the pick?

Step 1: Write 5 buyer questions without your brand name

Write the questions a buyer asks before they know you exist. Good ones look like this:

  • "What's the best [category] for a small business?"
  • "What's the best [category] for [specific use case]?"
  • "Cheapest [category] that does [key feature]?"
  • "[Competitor] alternatives?"
  • "Is [category] worth it for a team of 5?"

Leave your brand name out. If you ask "Is [your brand] good?", you've told the model who to talk about, and the answer stops telling you whether a buyer would find you. We saw this in our own testing: when a prompt named specific products, ChatGPT's answer opened with "Between the products you named..." and stayed inside that list.

Step 2: Use a clean chat with search on

Your account history can change the answer. OpenAI says ChatGPT search rewrites your question into one or more search queries, and "if memory is enabled, ChatGPT may use relevant saved memories when rewriting a search query." It also uses an approximate location from your IP address (OpenAI Help: Searching the web with ChatGPT).

To get closer to what a stranger sees:

  • Use a temporary chat, or a logged-out browser window. OpenAI says people who are not signed in can also use web search.
  • Turn on Search (in the tools menu, or type / and pick Search) so the answer pulls current pages instead of only the model's memory.
  • Don't add details about your own company to the chat.

Step 3: Ask ChatGPT, then Perplexity

Paste each question into ChatGPT, then into Perplexity. Two engines matter because they don't always agree. In our runs, they picked different winners for two of the four categories we tested (details below). If you only check one, you'll see half the picture.

Step 4: Log five things per answer

A plain spreadsheet works. One row per answer:

QuestionEngineNamed?Picked?Who was pickedPages cited
Best [category] for a small businessChatGPTYes, as an alternativeNoCompetitor Acompetitor-a.com/pricing

The "pages cited" column is the one most people skip, and it's the most useful. It shows you which pages the engine trusted for the answer: a competitor's pricing page, a review site, a "best tools" list. Those are the pages you need to be on, or to beat.

Open the cited pages, too. OpenAI's own help page says "search results and citations can be incomplete, outdated, or incorrect" and tells users to open a cited source to check it supports the answer. A citation that misdescribes you is worth knowing about.

Step 5: Run it more than once

One run is a snapshot, not a verdict. Answers can change between runs, between days, and between people. Re-run the same five questions on another day before you draw conclusions, and keep the old rows so you can compare. Weekly is a sensible rhythm if you're making changes and want to see whether they land.

What our own runs showed (one run each)

On Oct 27, 2026, we asked ChatGPT and Perplexity four buyer questions through their official APIs with web search on. The prompt named no brands: "A buyer asked: [question]... Name specific products and say which one you would tell them to use." Each was a single run, so treat these as examples of what you'll see, not rankings. API answers can also differ from what you'd get in the chat app.

  • Email marketing for a small business: ChatGPT picked Constant Contact. Perplexity picked Brevo. Mailchimp was named by both, but only as an alternative. Named, not picked.
  • Project management for a small team: ChatGPT picked Asana. Perplexity picked Trello.
  • AI coding assistant: both picked GitHub Copilot. The runner-up differed: Cursor in ChatGPT, Claude Code in Perplexity.
  • CRM for a small business: both picked HubSpot. Perplexity's search results included Pipedrive's pricing page, but the answer didn't name Pipedrive. Being found by the search isn't the same as being in the answer.

Two more patterns stood out:

  • Pricing pages got cited a lot. 4 of the 8 answers linked to a pricing page. For project management, both answers cited only pricing pages (Asana and monday.com in ChatGPT; Trello and Asana in Perplexity).
  • Brands' own pages got cited. ChatGPT's CRM answer linked HubSpot's CRM product page, and its AI coding answer linked GitHub's Copilot page.

The takeaway for your own check: look at which kind of page gets cited in your category. If it's pricing pages, make sure your price is on a page that can be read and quoted.

If you're not showing up: three things to check first

1. Can the search crawlers reach your site? ChatGPT search uses OAI-SearchBot. OpenAI says sites that opt out of OAI-SearchBot won't be shown in ChatGPT search answers, and that GPTBot (used for training) is a separate setting you can block without blocking search (OpenAI crawlers). Perplexity uses PerplexityBot to surface and link sites in its search results (Perplexity crawlers). Check your robots.txt and your CDN or firewall rules for both. Both companies say robots.txt changes can take about 24 hours to apply.

2. Is the page the answer needs actually there? If competitors get cited for their pricing page and you don't publish prices, there's nothing of yours to cite. The same goes for comparison and "alternatives" pages.

3. Are you measuring the visits? ChatGPT adds utm_source=chatgpt.com to referral links it sends, so you can filter for it in your analytics (OpenAI Publishers FAQ). It won't catch everyone. People who read an answer and type your URL in later won't carry the tag, so a "how did you hear about us?" question at signup helps fill the gap.

One honest caveat: OpenAI says "placement is not guaranteed." No checklist makes an engine pick you. What the check gives you is a clear view of where you stand and which pages are winning, so you know what to fix first.

Skip the spreadsheet: run the free check

If you'd rather not copy and paste 10 answers, Mentioned runs a related check for you. It works a little differently from the manual method above: the free check asks five buyer questions in ChatGPT and Perplexity (through their official APIs, with web search) and tells each engine that the buyer is comparing your product and domain with up to two competitors you choose. So it doesn't test whether a buyer would find you cold. It tests whether, when you're already on the shortlist, the AI would actually recommend you over those competitors. It shows whether you were mentioned or recommended, where you ranked against them, and which pages were cited. You also get three fixes to start with. It's one run, it's free, and there's no card. Use both: the manual method tells you whether you get found, and the free check tells you whether you win the comparison.

Run the free check at mentionedhq.com

Sources